Omega-3 long-chain polyunsaturated fatty acids (LC-PUFA) are now widely recognized as important compounds for human and animal health. These fatty acids may be obtained from dietary sources or by conversion of linoleic (LA, 18:2ω6) or α-linolenic (ALA, 18:3ω3) fatty acids, both of which are regarded as essential fatty acids in the human diet. While humans and many other vertebrate animals are able to convert LA or ALA, obtained from plant sources to C22 they carry out this conversion at a very low rate. Moreover, most modern societies have imbalanced diets in which at least 90% of polyunsaturated fatty acids (PUFA) are of the ω6 fatty acids, instead of the 4:1 ratio or less for ω6:ω3 fatty acids that is regarded as ideal (Trautwein, 2001). The immediate dietary source of LC-PUFAs such as eicosapentaenoic acid (EPA, 20:5ω3), docosapentaenoic acid (DPA) and docosahexaenoic acid (DHA, 22:6ω3) for humans is mostly from fish or fish oil. Health professionals have therefore recommended the regular inclusion of fish containing significant levels of LC-PUFA into the human diet. Increasingly, fish-derived LC-PUFA oils are being incorporated into food products and in infant formula, for example. However, due to a decline in global and national fisheries, alternative sources of these beneficial health-enhancing oils are needed.
Flowering plants, in contrast to animals, lack the capacity to synthesise polyunsaturated fatty acids with chain lengths longer than 18 carbons. In particular, crop and horticultural plants along with other angiosperms do not have the enzymes needed to synthesize the longer chain ω3 fatty acids such as EPA, docosapentaenoic acid (DPA, 22:5ω3) and DHA that are derived from ALA. An important goal in plant biotechnology is therefore the engineering of crop plants which produce substantial quantities of LC-PUFA, thus providing an alternative source of these compounds.
LC-PUFA Biosynthesis Pathways
Biosynthesis of LC-PUFAs in organisms such as microalgae, mosses and fungi usually occurs as a series of oxygen-dependent desaturation and elongation reactions (FIG. 1). The most common pathway that produces EPA in these organisms includes a Δ6-desaturation, Δ6-elongation and Δ5-desaturation (termed the Δ6-desaturation pathway) whilst a less common pathway uses a Δ9-elongation, Δ8-desaturation and Δ5-desaturation (termed the Δ9-desaturation pathway). These consecutive desaturation and elongation reactions can begin with either the ω6 fatty acid substrate LA, shown schematically as the upper left part of FIG. 1 (ω6) or the ω3 substrate ALA through to EPA, shown as the lower right part of FIG. 1 (ω3). If the initial Δ6-desaturation is performed on the ω6 substrate LA, the LC-PUFA product of the series of three enzymes will be the ω6 fatty acid ARA, LC-PUFA synthesising organisms may convert ω6 fatty acids to ω3 fatty acids using an ω3-desaturase, shown as the Δ17-desaturase step in FIG. 1 for conversion of arachidonic acid (ARA, 20:4ω6) to EPA. Some members of the ω3-desaturase family can act on a variety of substrates ranging from LA to ARA. Plant ω3-desaturases often specifically catalyse the Δ15-desaturation of LA to ALA, while fungal and yeast ω3-desaturases may be specific for the Δ17-desaturation of ARA to EPA (Pereira et al., 2004a; Zank et al., 2005). Some reports suggest that non-specific ω3-desaturases may exist which can convert a wide variety of ω6 substrates to their corresponding ω3 products (Zhang et al., 2008).
The conversion of EPA to DHA in these organisms occurs by a Δ5-elongation of EPA to produce DPA, followed by a Δ4-desaturation to produce DHA (FIG. 1). In contrast, mammals use the so-called “Sprecher” pathway which converts DPA to DHA by three separate reactions that are independent of a Δ4-desaturase (Sprecher et al., 1995).
The front-end desaturases generally found in plants, mosses, microalgae, and lower animals such as Caenorhabditis elegans predominantly accept fatty acid substrates esterified to the sn-2 position of a phosphatidylcholine (PC) substrate. These desaturases are therefore known as acyl-PC, lipid-linked, front-end desaturases (Domergue et al., 2003). In contrast, higher animal front-end desaturases generally accept acyl-CoA substrates where the fatty acid substrate is linked to CoA rather than PC (Domergue et al., 2005). Some microalgal desaturases and one plant desaturase are known to use fatty acid substrates esterified to CoA (Table 2).
Each PUFA elongation reaction consists of four steps catalysed by a multi-component protein complex: first, a condensation reaction results in the addition of a 2C unit from malonyl-CoA to the fatty acid, resulting in the formation of a β-ketoacyl intermediate. This is then reduced by NADPH, followed by a dehydration to yield an enoyl intermediate. This intermediate is finally reduced a second time to produce the elongated fatty acid. It is generally thought that the condensation step of these four reactions is substrate specific whilst the other steps are not. In practice, this means that native plant elongation machinery is capable of elongating PUPA providing that the condensation enzyme (typically called an ‘elongase’) specific to the PUFA is introduced, although the efficiency of the native plant elongation machinery in elongating the non-native PUFA substrates may be low. In 2007 the identification and characterisation of the yeast elongation cycle dehydratase was published (Denic and Weissman, 2007).
PUFA desaturation in plants, mosses and microalgae naturally occurs to fatty acid substrates predominantly in the acyl-PC pool whilst elongation occurs to substrates in the acyl-CoA pool. Transfer of fatty acids from acyl-PC molecules to a CoA carrier is performed by phospholipases (PLAs) whilst the transfer of acyl-CoA fatty acids to a PC carrier is performed by lysophosphatidyl-choline acyltransferases (LPCATs) (FIG. 9) (Singh et al., 2005).
Engineered Production of LC-PUFA
Most LC-PUFA metabolic engineering has been performed using the aerobic Δ6-desaturation/elongation pathway. The biosynthesis of γ-linolenic acid (GLA, 18:3ω6) in tobacco was first reported in 1996 using a Δ6-desaturase from the cyanobacterium Synechocystis (Reddy and Thomas, 196). More recently, GLA has been produced in crop plants such as safflower (73% GLA in seedoil, WO 2006/127789) and soybean (28% GLA; Sato et al., 2004). The production of LC-PUFA such as EPA and DHA involves more complicated engineering due to the increased number of desaturation and elongation steps involved. EPA production in a land plant was first reported by Qi et al. (2004) who introduced genes encoding a Δ9-elongase from Isochrysis galbana, a Δ5-desaturase from Euglena gracilis and a Δ5-desaturase from Mortierella alpina into Arabidopsis yielding up to 3% EPA. This work was followed by Abbadi et at (2004) who reported the production of up to 0.8% EPA in flax seed using genes encoding a Δ6-desaturase and Δ6-elongase from Physcomitrella patens and a Δ5-desaturase from Phaeodactylum tricornutum. 
The first report of DHA production was in WO 04/017467 where the production of 3% DHA in soybean embryos is described, but not seed, by introducing genes encoding the Saprolegnia diclina Δ6-desaturase, Mortierella alpina Δ6-desaturase, Mortierella alpina Δ5-desaturase, Saprolegnia diclina Δ4-desaturase, Saprolegnia diclina Δ17-desaturase, Mortierella alpina Δ6-elongase and Pavlova lutheri Δ5-elongase. The maximal EPA level in embryos also producing DHA was 19.6%, indicating that the efficiency of conversion of EPA to DHA was poor (WO 2004/071467). This finding was similar to that published by Robert et al. (2005), where the flux from EPA to DHA was low, with the production of 3% EPA and 0.5% DHA in Arabidopsis using the Danio rerio Δ5/6-desaturase, the Caenorhabditis elegans Δ6-elongase, and the Pavlova salina Δ5-elongase and Δ4-desaturase. Also in 2005, Wu et al. published the production of 25% ARA, 15% EPA, and 1.5% DHA in Brassica juncea using the Pythium irregulare Δ6-desaturase, a Thraustochytrid Δ5-desaturase, the Physcomitrella patens Δ6-elongase, the Calendula officanalis Δ12-desaturase, a Thraustochytrid Δ5-elongase, the Phytophthora infestans Δ17-desaturase, the Oncorhyncus mykiss LC-PUFA elongase, a Thraustochytrid Δ4-desaturase and a Thraustochytrid LPCAT (Wu et al., 2005). Summaries of efforts to produce oil-seed crops which synthesize ω3 LC-PUFAs is provided in Venegas-Caleron et al. (2010) and Ruiz-Lopez et al. (2012). As indicated by Ruiz-Lopez et al. (2012), results obtained to date for the production of DHA in transgenic plants has been no where near the levels seen in fish oils. More recently, Petrie et al (2012) reported the production of about 15% DHA in Arabidopsis thaliana seeds, and WO2013/185184 reported the production of certain seedoils having between 7% and 20% DHA.
To date, recombinant cells, such as recombinant plant cells, producing LC-PUFA have a propensity to esterify the LC-PUFA at the sn-1 and/or sn-3 position of triacylglycerols (TAG) which limits the total amount of LC-PUFA which can be found as TAG in these cells. There therefore remains a need for more efficient production of TAG where LC-PUFA is esterified at the sn-2 position in recombinant cells, in particular in seeds of oilseed plants.